Artificial Intelligence Models for Precision Prediction and Treatment of Prostate Cancer

NCT06662708 | Not Yet Recruiting DMC

• 1 other identifier: Shao Pengfei
• Study Type: interventional
• Target: 200
• Locations: 1 country
• Sites: 1

Brief Summary

The aim of this clinical trial is whether artificial intelligence models can be used for accurate clinical preoperative diagnosis and postoperative diagnosis of pathological findings, and will also measure the accuracy of the predictions made by the artificial intelligence models.The main target questions addressed by the model building are:

1. 1.whether the AI model can learn from preoperative MRI and postoperative Whole Slide Images so as to accurately predict information such as benignness or malignancy, aggressiveness, grading, subtypes, genes, etc. for participants suspected of having prostate cancer preoperatively/puncturally.
2. 2.whether the AI model is capable of learning postoperative macropathology slides to enable outcome diagnosis of surgical pathology slides in new participants.
3. 3.complete an MRI examination and have their MRI images analysed by the established AI model to make an accurate diagnosis of them.
4. 4.Based on the diagnosis, if prostate cancer is predicted, they will undergo radical prostate cancer surgery and refine their surgical pathology.

Conditions

Prostate Cancer; Prostate Intraductal Carcinoma; Prostate Cancer Aggressiveness; Prostate Cancer Stage; Pathology

Keywords

Articicial Intelligence; Whole Slide Image; mp-MRI; Prostate Cancer; Prediction Model

Outcome Measures

Primary Outcomes (3)

• Prediction of postradical prostate cancer pathology after radical prostatectomy using the 'AUC' comprehensive assessment model

  'AUC' refers to the area under the ROC (Receiver Operating Characteristic) curve, which indicates the performance of the model in predicting immunohistochemistry-related pathological information of prostate cancer after surgery, and the AUC ranges from 0-1, with the larger value indicating the better prediction effect.

  From subject enrolment to initial post-surgery, usually 30-90 days.

• Predicting the performance of post-radical pathology by the 'AUC' comprehensive assessment model

  'AUC' refers to the area under the ROC (Receiver Operating Characteristic) curve, indicating the level of performance of the model in predicting prostate cancer in the preoperative period, with AUC ranging from 0-1, with larger values indicating better prediction results.

  From subject enrolment to initial post-surgery, usually 30-90 days.

• 'F1 Score' to assess performance of preoperative 3D modelling

  A reconciled average of the preoperative 3D modelling precision and recall assessed through the 'F1 score', which represents the match to the real situation.

  From subject enrolment to initial post-surgery/puncture recovery, usually 30-90 days.

Secondary Outcomes (2)

• Assess the amount of cost difference between the predictive model and the clinical approach by "economic cost savings"

  From subject enrolment to initial post-surgery/puncture recovery, usually 30-90 days.

• "Diagnostic Time" evaluate the time taken to predict immunohistochemistry-related pathology in the postoperative period.

  From subject enrolment to initial post-surgery/puncture recovery, usually 30-90 days.

Study Arms (2)

Experimental group

EXPERIMENTAL

This group of patients will receive predictions assisted by artificial intelligence models.

Diagnostic Test: Accurate Prediction Artificial Intelligence Models

Control Group

NO INTERVENTION

This group of patients will not receive predictions assisted by artificial intelligence models.

Interventions

Accurate Prediction Artificial Intelligence Models DIAGNOSTIC_TEST

Diagnostic Test: Accurate Prediction Artificial Intelligence Models Post-operative pathology, precise pre-operative diagnosis (including benign and malignant, invasive, grading, subtypes) or 3D lesion modelling will be predicted based on the AI predictive model in response to the information provided

Experimental group

Eligibility Criteria

• Age: 30 Years+
• Sex: male
• Healthy Volunteers: Yes
• Age Groups: Adult (18-64), Older Adult (65+)

You may qualify if:

• Patients with suspected PCa (elevated PSA or suspicious positive lesions on ultrasound or MRI results);

You may not qualify if:

• Previous treatment of the prostate in any form, including surgery, radiotherapy/chemotherapy, endocrine therapy, targeted therapy and immunotherapy;
• Patients with any item missing from the baseline clinical and pathological information;
• Patients with a history of other malignancies, serious comorbidities or other health problems;
• Unable to provide/sign an informed consent form;
• Patients who, in the judgement of the investigator, are deemed unfit to participate in this clinical trial;

Sponsors & Collaborators

• Shao Pengfei: lead
• Institute of Automation, Chinese Academy of Sciences: collaborator

Study Sites (1)

The First Affiliated Hospital of Nanjing Medical University (Jiangsu Provincial People's Hospital)

Nanjing, Jiangsu, 210036, China

Location

MeSH Terms

Conditions

Prostatic Neoplasms; Prostate Cancer, Hereditary, 7

Condition Hierarchy (Ancestors)

Genital Neoplasms, Male; Urogenital Neoplasms; Neoplasms by Site; Neoplasms; Genital Diseases, Male; Genital Diseases; Urogenital Diseases; Prostatic Diseases; Male Urogenital Diseases

Central Study Contacts

Pengfei Shao, Professor

CONTACT

13851925825; spf032@hotmail.com

Pan Zang, Postgraduate

CONTACT

18914730216; 896381164@qq.com

Study Design

• Study Type: interventional
• Phase: not applicable
• Allocation: RANDOMIZED
• Masking: TRIPLE
• Who Masked: PARTICIPANT, CARE PROVIDER, OUTCOMES ASSESSOR
• Purpose: DIAGNOSTIC
• Intervention Model: PARALLEL
• Sponsor Type: OTHER
• Responsible Party: SPONSOR INVESTIGATOR
• PI Title: Chief physician

Model Details: Inclusion of enrolled patients in an artificial intelligence predictive model that predicts postoperative pathology, precise preoperative diagnosis (including benign and malignant, invasive, grading, and subtypes) or 3D lesion modelling based on the information provided

Study Record Dates

• First Submitted: August 13, 2024
• First Posted: October 29, 2024
• Study Start: December 1, 2024
• Primary Completion (Estimated): January 1, 2030
• Study Completion (Estimated): December 31, 2030
• Last Updated: October 29, 2024

Record last verified: 2024-10

Locations

CN (1)